I placed 120 units in the Paypal inventory for the 10pm sale, although Paypal oversold again so a total of 135 units were sold, although one was cancelled, so it’s 134 really.
The oversold inventory expired at 22:00:49.
We had a total of 140 available, of which 60 were built by yours truly and 80 by my new manufacturer. Those 80 units arrived from the manufacturer on Friday at 2pm. I had tested all of them by 6pm. 25 had faults, and I’d fixed 12 of those on the fly during the testing as they were simple faults. The remaining 13 faults were fixed on Saturday morning. I then set to on the HPF rework on all those 80 units on Saturday afternoon. This was completed by Saturday night, however they were not tested until this morning (Sunday). Complete testing and checkout of all 140 units was completed today (Sunday) at 2pm. There were no write offs.
From Saturday morning until Sunday evening, I had one relative full time performing dispatch, and on Saturday afternoon I had another passing relative preparing the enclosures (labelling, drilling etc). They’re young, and are not easily offended by the offer of cold hard cash. Normally over the weekend sales I do the labelling with graciously accepted help from the OH, and help out with dispatch, but the later arrival of the units from the manufacturer meant I was busy getting the hardware itself ready.
Each unit is system tested three times. Firstly immediately after construction (when the bootloader is also installed), secondly after the HPF rework and thirdly after placement in the enclosure. When rework is done in large batches such as in this last batch, the second and third tests are performed at the same time by testing the unit out side of the enclosure and if it tests well it is immediately placed in the enclosure and tested in situ there.
I’ve found in tests after the HPF rework has been installed often results in having to replace the LNA as its sensitivity seems to be impaired by a dB or two. This appears to be due to the thermal cycling. This occurs in abut 10% of boards.
In addition, the tuner chips need replacement about 5% of the time due to I/Q imbalance being out of spec.
I found an exceptional number of PCB faults too on these latest boards which is very disappointing. About half a dozen breaks in tracks that must have been caused by faulty photographics when looked at under the microcscope. Unless you’ve seen them before, debugging these is a very lengthy process often with numerous blind alleys.
There also seems to be a problem with some 1nF caps I am using that seem to frequently suffer cracks. I will change the manufacturer for the next runs.
Here is the feedback to the manufacturer on the 25 or so units that failed.
xx1 Crystal 1000ppm HF, also may be coincidence, looks like may not have been soldered properly on one lug, fixed
xx2 USB connector plastic deformed, I assume something happened during reflow, it would not physically mate, replaced connector, fixed
xx0 IC8 pin 6 lifted, fixed
xx3 high phase noise, intermittent, replaced IC3, fixed
xx8 Clock low p-p swing, replaced IC2, fixed
xx9 deaf, replaced C1 (cracked), fixed
xx6.1 I/Q imbalance 20dB, replaced IC3, fixed
xx6.2 deaf, replaced IC4 and C1 (cracked), fixed
xx7.1 stuck in bootloader, replaced IC10 3.3A reg, fixed
xx7.2 deaf, replaced IC4, fixed
xx0 codec data track broken (PCB fault), fixed
xx2 stuck in bootloader, track broken pin 32 IC3 (PCB fault), fixed
xx3 codec data track broken (PCB fault), fixed
xx3 I/Q imbalance 10dB, replaced IC3, fixed
xx4 USB connector D- line connection lifted, fixed
xx3 I/Q imbalance 20dB, replaced IC3, fixed
xx4 I/Q imbalance 20dB, replaced IC3, fixed
xx6 I/Q imbalance 20dB, replaced IC3, fixed
xx8 deaf, replaced IC3 and C1 (cracked), fixed
xx4 codec data track broken (PCB fault), fixed
xx5 stuck in bootloader, IC2 solder bridge pins 7 & 8, fixed
xx8 I/Q imbalance 20dB, replaced IC3, fixed
xx2 stuck in bootloader, solder bridge on IC2 pins 17 & 18, fixed
xx7 totally deaf, lifted PCB track IC3 pin 8, fixed
xx9 stuck in bootloader, IC2 incorrectly oriented, fixed
I had a cunning plan; in light of the response at the last Roundtable, how about bring some to the next one, and sell them at the event.
I am sure that MMRT team would be delighted, and you are sure of a complete sell out,…?
Looks like I was one of the lucky ones last Friday. I just got an email notice that a package is on its way 🙂
I was also one of the lucky last friday but no follow-up email. Still waiting 🙂
I have one Philippe in my manifest. All items were posted today (Monday 7 February).
Mine is also on the way !
The address has a minor fault. Should be no problem, however as the postal code is right; I tried the PayPal help but this lead me to NowhereNoplaces…… It’s impossible to alert them that the address is faulty! They write “unconfirmed address” but there is no way to confirm or correct it. I hope Mr. Murphy will be busy elsewhere this week!
73 Howard, many thanks for a very interesting dongle!
Augusto i2jjr / 9h3jr
I sent it to the address in the separate email that you sent.
Thanks Howard ! Greatly appreciated.
1- Have you considered reserving a few units and sending them to SDR software authors free of charge? It’d probably go a long way toward getting them supported natively within some of the more popular SDR apps like SDR-Radio, wrplus, CuteSDR, Winrad, etc
2- Is it typical to have so many failed units come back from a manufacturer? Why aren’t they doing the testing/repairing?
Yes, I did have a conversation with one person but I have been unable to get their delivery address.
Regarding the testing/repairing, please remember that I am not an expert in this and it is a big learning experience for me. The level of failure was high, but far better than the previous manufacturer. What was strange was that there were many different failure modes. With amy manufacturing process it will take some tweaking. I think a lot of people think it’s simply a matter of turning on a tap and everything works.
Please also remember that I do perform a number of fairly rigourous checks, and for example I still rework boards with I/Q imbalances that could easily be corrected in software, and fix boards that are slightly outside my sensitivity criteria.
The manufacturer is not doing the testing or repairing yet because we haven’t quite got the manufacturing process right yet to allow us to characterise common faults. It is my intention to get them to do the functional tests, but they will also need the right equipment like comms test sets to do this and appropriate training.
The biggest disappointment was the PCBs which I deliberately used the previous company as they had been very good on the previous orders of exactly the same boards. On this second occasion, they must have left the artwork unprotected as there were repeated broken tracks. In addition there were routing problems with a few boards.